Safe Cracking with Robots

What's in the box?! Opening a safe with a little intelligence and a little brute force.

Favorited Favorite 3

UPDATE: SUCCESS! We managed to open the safe in under an hour. Believe me, we’re as shocked as anyone. We’re now going to tabulate the winners, and work on our full write-up of how we did what we did. Thanks for joining us!

Safe cracked with advanced autodialer

The live stream is done but you can re-watch the stream if you’d like. The magic moment occurs at 45:20 but start around 44:30 to get the full scope of what’s going on.

I have been infatuated with locks and mechanisms from a very early age. For my 7th birthday, my grandparents gave me a box of locks and keys; I promptly drove my parents insane as I ‘locked’ things around the house. Later, I signed up for all sorts of mail order ‘Train to be a Locksmith’ publications in an attempt to learn what was then the taboo black art of lockpicking.

Fast forward a few dozen years, and lock picking tools and how-to videos are cheap and prevalent. I have picked my way through many lock types and enjoyed teaching friends the meaning behind ‘locks keep honest people honest’ (household locks are surprisingly easy to pick, they just keep out honest people). In a similar vein, I have wanted to learn how to crack safes like they do in the movies. Yes, yes, we know it’s fake. But for every bad Hollywood scenario there is usually a grain of legitimacy buried in there somewhere. Safe cracking is a profession that’s been around since safes were invented. I wanted to know how much work and how viable safe cracking could be.

My wife is amazing; for Christmas this year, she gave me a fire safe she purchased off of Craigslist. It was super cheap because the seller didn’t have the combination. Best present ever.

Step 1) Get a safe that hasn’t been opened.

Step 2) Deploy robot army.

Common home combination fire safe

My safe is the very regular home-brand that is not particularly advanced but secure enough to deter crimes of convenience. If you know anything about ‘picking’ safes, you do it by feel. Manufacturers know this, so this safe has a ratchet built into the dial that fairly effectively thwarts the ability to listen or feel with a few additional safety features built-in as well. We’ll cover in detail how we built the safe cracking robot and all the lessons we’ve learned shortly after the safe is cracked. Now, let’s talk contest…

$200 in Prizes

We’re going to live stream the cracking of our safe starting at 10AM MDT on Wednesday, March 29th. It’s going to be a cross between watching a cliff hanger and paint drying. We’re only about 60% sure we’ll get the safe open. Who knows how long it will take? Comment on this post with your guess of how much time you believe it will take us to open the safe in the form hh:mm:ss. The closest guess wins a $100 gift cert to SparkFun. Additionally, you are encouraged to guess the combination in the format AA/BB/CC. The closest combination guess will also win $100. Price-Is-Right rules apply: closest without going over. We’ll use your time and combination guesses to break any ties if they arise. Comments and entries close at 9:59AM Wednesday before the robot powers up.

Safe cracking robot front

Safe cracking robot top

Let me be clear, the robot theoretically works, but we have not yet opened the safe. We’re pretty sure there’s nothing in the safe. We’ve moved it a few times. Nothing rattles around, but who knows? Maybe there’s a bar of gold inside! Or a ham sandwich. We don’t know, but we’re going to find out.


Comments 41 comments

  • barryjh / last year / 3

    Well at least you’re not expecting anything to be inside. When Geraldo Rivera opened a safe live, it didn’t end well for him :)

    This is tough because you didn’t give much information as to how quickly you can try a combination or your method used. Even at 1 second per combination try, it would take over 11 days to try all combinations. So you must be doing something with listening, and then trying a combination based on where on the dial particular sounds showed up, after filtering out the ratchet sound of course. If its by listening, then it seems like a few listen cycles and then try the combination and see if it works. If it doesn’t not sure what step 2 is, perhaps changing some filtering parameters and listening again? If so seems like it would be pretty quick or never unless the algorithm can learn from its mistakes.

    I’m going to go with quick or never.

    crack time: 00:07:30, combo 26, 33, 92

    • Nate / last year / 2

      Hey Barry! We’re being intentionally coy :) If we described everything we’re doing, what would be the fun in that? The full write up will follow. Again, we’re confident but there’s always that worry if we’ve got the robot functioning correctly. Won’t know until we really let it rip.

      • barryjh / last year / 1

        I know it was intentional but I had to bitch anyway since the possible methods are numerous. I’ll be interested to see the method you went with and the reasons you selected the method. If it does include filtering sampled data I’ll be interested to see if you designed a FIR using TFilter :)

  • Time: 6:55:20 Combo: 55/55/65

  • 13 min, 56,9,37 combo

  • Time: 26:15:34, Combo 87/44/95

  • 1,000,000 possibilities at 1 try a second, a million seconds will be about 1/30 of a year or 11.574 or so days. Lets say the evil robot doesn’t try even numbers, the combo is the average all the combos posted thus far and then account for how long it might take to get there…Time: 58:45:59 Combo: 39/45/43

  • mikenon / last year / 1

    Time: 01:24:00 Combo: 08/10/12

  • Time 00:00:01 Combo 35/49/86

  • Time 3:31:41 15/ 67/47

  • GI_J0E / last year / 1

    Time: 84:46:23 Combo: 76/42/21

  • 15:34:00 6/24/92

  • Time 08:52:00 Combo: 89/12/22

  • Bob6 / last year / 1

    Time: 01:17:33 Combo: 26/41/17

  • Zap / last year / 1

    Time: 138:53:20 Combo: 51/50/51

  • Time: 07:38:17 Combo: 54:91:10

  • tycen / last year / 1

    Time 00:05:32 Combo 89/26/49

  • Time 06:30:00 combination 50/75/50

  • 28:15:32 for the time and 38/12/16 for the combo.

  • bdin / last year / 1

    Time: 01:23:33 Combo: 36/18/72

  • Time: 14:52:34 Combo: 56 / 74 / 68

    Very much interested to read the full write up of this.

  • Smok / last year / 1

    Looks like it is 1 000 000 combinations. Assuming 1 second per combination and just assuming the right combination would be at 75%, it’s 750 000 seconds or Time 280:00:00. Combination 75/99/99. I hope you are going to try them in random order, not sequentially. That would be more fun.

  • I’ll go with 18 hours 32 minutes, with a combination of 73/05/48. Lock picking is a hobby of an uncle of mine, but I have no idea about safes. While I no he has no time or ability at work to watch, I know he’s going to be interested.

  • Time, 1:00:00 Combination, 46/68/10

    Happy cracking!

  • Time: 00:00:30 Combo: 78/25/17

  • Time: 00:02:00 Combo: 10/05/63

  • 00:30:05 15/35/55 current sensor… sounds good ^_^

  • Time 04:12:57 Combo 45/94/03

  • 12:34:56 12/34/56

  • 00:03:14 14/06/03

  • Time 02:29:08 Combo 21/77/36

  • l0gikG8 / last year / 1

    Time: 00:12:00, combo: 42/65/12

  • Almost no information to base a guess upon, might as well try. 2 days 7 hr 23 mins – combo if not the factory combo (!) 88 - 27 - 61

  • dude124 / last year / 1

    Time: 05:44:23 Combo: 66/106/28

    • We didn’t point it out, but the number range on the dial is 0-99, so you may want to edit your second number.

  • saccade / last year / 1

    I’m a little queasy about the server being strong enough to actually pull the latch open.

    So many questions to field an accurate guess: How many tries per minute can the robot do? Have you optimized the algorithm for trying combinations?

    I’m guessing the robot isn’t super fast. 12:35:00. 12-80-20

  • It is always best to consider the following: Look up the default combination for the lock type. Try that. If it doesn’t work, try variations on typical combos: 10-20-30, 36-24-36, 50-0, 50-25-50, 25-50-25. etc. That may get you open long before the autodialer would. lockmasters.com sells professional equipment to do this type of work Dialing it in mechanically is a fun exercise, most locksmiths would look in a manual on where to cut to disable the bolt mechanism but that usually leaves the safe unusable unless you consider using a small hole and a borescope, more finesse there. If things are set to be autodialed, I say 38 hours, 20-40-60.

  • GCF / last year / 1

    Time - 10:42:00 Combo - 12/76/53

  • Ruba / last year / 1

    Time: 03:10:45, combo: 19/5/45

  • Time: 04:26:32 Combo: 25/50/25

  • Time: 01:13:29 combo: 17,3,29